Matrix Rep. Same basics as introduced already.
Convenient method of working with vectors.
Superposition Complete set of vectors can be used to
express any other vector.
Complete set of N vectors can form other complete sets of N vectors.
Can find set of vectors for Hermitian operator satisfying
Eigenvectors and eigenvalues
Matrix method Find superposition of basis states that are
eigenstates of particular operator. Get eigenvalues.
. A u u o =
Copyright – Michael D. Fayer, 2007
Orthonormal basis set in N dimensional vector space
{ }
j
e basis vectors
1
N
j
j
j
x x e
=
=
¿
j
j
x e x =
Any vector can be written as
with
To get this, project out
from
sum over all .
j j
j
j j j
j
e e x
e
x e e e x =
Copyright – Michael D. Fayer, 2007
Operator equation
y A x =
1 1
N N
j j
j j
j j
y e A x e
= =
=
¿ ¿
1
N
j
j
j
x A e
=
=
¿
Substituting the series in terms of bases vectors.
i
e
1
N
i j
i j
j
y e A e x
=
=
¿
Left mult. by
i j
e A e
{ }
and the basis set .
j
A e
The N
2
scalar products
are completely determined by
N values of j ; N for each y
i
Copyright – Michael D. Fayer, 2007
{ }
j
e
Writing
i j
ij
a e A e =
1
N
i ij j
j
y a x
=
=
¿ 1 2 , , j N =
Matrix elements of A in the basis
gives for the linear transformation
Know the a
ij
because we know A and
{ }
j
e
In terms of the vector representatives
 
1 2
, ,
N
x x x x =
 
1 2
, ,
N
y y y y =
(Set of numbers, gives you vector
when basis is known.)
y Ax =
The set of N linear algebraic equations can be written as
double underline means matrix
ˆ ˆ ˆ
7 5 4
[7, 5, 4]
Q x y z = + +
vector
vector representative,
must know basis
Copyright – Michael D. Fayer, 2007
A
( )
11 12 1
21 22 2
1 2
N
N
ij
N N NN
a a a
a a a
A a
a a a
(
(
(
= =
(
(
¸ ¸
array of coefficients  matrix
The a
ij
are the elements of the matrix .
A
A
The product of matrix and vector representative x
is a new vector representative y with components
1
N
i ij j
j
y a x
=
=
¿
Copyright – Michael D. Fayer, 2007
Matrix Properties, Definitions, and Rules
A B
A B =
Two matrices, and are equal
if a
ij
= b
ij
.
1 0 0
0 1 0
1
0 0 1
ij
o
(
(
(
= =
(
(
¸ ¸
The unit matrix
ones down
principal diagonal
1
N
i ij j i
j
y x x o
=
= =
¿
1 y x x = =
Gives identity transformation
Corresponds to
0 0 0
0 0 0
0
0 0 0
(
(
(
=
(
(
¸ ¸
0 0 x=
The zero matrix
Copyright – Michael D. Fayer, 2007
y A x = z B y =
z BA x =
Matrix multiplication
Consider
1
N
k ki i
i
z b y
=
=
¿
z B y =
z By BAx Cx = = =
Using the same basis for both transformations
matrix ( )
ki
B b = =
C BA =
1
N
kj ki ij
i
c b a
=
=
¿
has elements
Law of matrix multiplication
operator equations
Example
2 3 7 5 29 28
3 4 5 6 41 39
    
=
  
\ .\ . \ .
Copyright – Michael D. Fayer, 2007
Multiplication Associative
( ) ( )
B A C A BC =
B A BA =
Multiplication NOT Commutative
A B C o  + =
ij ij ij
c a b o  = +
Matrix addition and multiplication by complex number
A
1
A
÷
1 1
1 AA A A
÷ ÷
= =
A
Inverse of a matrix
inverse of
identity matrix
CT
1
0 If 0 is singular
A
A
A
A A A
÷
=
= =
transpose of cofactor matrix (matrix of signed minors)
determinant
Copyright – Michael D. Fayer, 2007
Reciprocal of Product
( )
1
1 1
AB B A
÷
÷ ÷
=
( )
ij
A a =
( )
ji
A a =
For matrix defined as
interchange rows and columns
Transpose
( )
*
*
ij
A a =
Complex Conjugate
complex conjugate of each element
( )
*
ji
A a
+
=
Hermitian Conjugate
complex conjugate transpose
Copyright – Michael D. Fayer, 2007
Rules
( )
AB BA = transpose of product is product of transposes in reverse order
    A A = determinant of transpose is determinant
* *
*
( ) AB A B =
complex conjugate of product is product of complex conjugates
*
*
    A A =
determinant of complex conjugate is
complex conjugate of determinant
( ) AB B A
+ +
+
= Hermitian conjugate of product is product of
Hermitian conjugates in reverse order
*
    A A
+
=
determinant of Hermitian conjugate is complex conjugate
of determinant
Copyright – Michael D. Fayer, 2007
Definitions
A A =
A A
+
=
*
A A =
*
A A = ÷
1
A A
÷ +
=
ij ij ij
a a o =
Symmetric
Hermitian
Real
Imaginary
Unitary
Diagonal
0 1 2
1 A A A A AA = = = · · ·
2
1
2!
A
A
e A = + + + · · ·
Powers of a matrix
Copyright – Michael D. Fayer, 2007
1
2
N
x
x
x
x
(
(
(
=
(
(
¸ ¸
Column vector representative one column matrix
y Ax =
1 11 12 1
2 21 22 2
1 2
N
N N N NN N
y a a a x
y a a x
y a a a x
( ( (
( ( (
( ( (
=
( ( (
( ( (
¸ ¸ ¸ ¸ ¸ ¸
then
vector representatives in particular basis
becomes
( )
1 2
,
N
x x x x =
y Ax =
y xA =
y Ax = y x A
+
+ +
=
row vector transpose of column vector
transpose
Hermitian conjugate
Copyright – Michael D. Fayer, 2007
Change of Basis
{ }
i
e
( )
, 1, 2,
i j
ij
e e i j N o = =
orthonormal basis
then
{ }
i
e
{ }
i
e
'
1
1, 2,
N
i k
ik
k
e u e i N
'
=
= =
¿
Superposition of can form N new vectors
linearly independent
a new basis
complex numbers
Copyright – Michael D. Fayer, 2007
j i
ij
e e o
' '
=
New Basis is Orthonormal
if the matrix
( )
ik
U u =
1 U U
+
=
1
U U
÷ +
=
coefficients in superposition
1
1, 2,
N
i k
ik
k
e u e i N
'
=
= =
¿
meets the condition
is unitary U
{ }
i
e
'
U
1 UU U U
+ +
= =
Important result. The new basis will be orthonormal
if , the transformation matrix, is unitary (see book
and Errata and Addenda ).
Copyright – Michael D. Fayer, 2007
{ }
e
'
{ }
e
Unitary transformation
substitutes orthonormal basis for orthonormal basis .
x
i
i
i
x x e =
¿
i
i
i
x x e
'
'
=
¿
x
Vector
Same vector – different basis.
vector – line in space (may be high dimensionality
abstract space)
written in terms of two basis sets
U
x
x U x
'
=
x U x
+
'
=
The unitary transformation can be used to change a vector representative
of in one orthonormal basis set to its vector representative in another
orthonormal basis set.
x – vector rep. in unprimed basis
x' – vector rep. in primed basis
change from unprimed to primed basis
change from primed to unprimed basis
Copyright – Michael D. Fayer, 2007
{ }
ˆ ˆ ˆ
, , x y z
y
x
z
s
ˆ ˆ ˆ
7 7 1 s x y z = + +
s
{ }
ˆ ˆ ˆ
, , x y z
7
7
1
s
(
(
=
(
(
¸ ¸
Example
Consider basis
Vector  line in real space.
In terms of basis
Vector representative in basis
Copyright – Michael D. Fayer, 2007
Change basis by rotating axis system 45° around .
ˆ
z
s
s Us
'
=
cos sin 0
sin cos 0
0 0 1
x y z
U
u u
u u
 

= ÷


\ .
2 / 2 2 / 2 0
2 / 2 2 / 2 0
0 0 1
U
 

= ÷ 


\ .
Can find the new representative of , s'
U is rotation matrix
For 45° rotation around z
Copyright – Michael D. Fayer, 2007
2 / 2 2 / 2 0
7 7 2
2 / 2 2 / 2 0 7 0
0 0 1 1 1
s
 
 
 



' = ÷ = 






\ .
\ .
\ .
7 2
0
1
s
(
(
'
=
(
(
¸ ¸
Then
vector representative of in basis s
{ }
e
'
Same vector but in new basis.
Properties unchanged.
( )
1/ 2
s s
1/ 2
1/ 2 1/ 2 1/ 2
*
( ) (49 49 1) (99) s s s s ( = · = + + =
¸ ¸
1/ 2
1/ 2 1/ 2 1/ 2
*
( ) (2 49 0 1) (99) s s s s
' ' ( = · = × + + =
¸ ¸
Example – length of vector
Copyright – Michael D. Fayer, 2007
Can go back and forth between representatives of a vector by x
x U x
'
=
x U x
+
'
=
change from unprimed
to primed basis
change from primed
to unprimed basis
x components of in different basis
Copyright – Michael D. Fayer, 2007
y A x =
Consider the linear transformation
operator equation
y Ax =
i ij j
j
y a x =
¿
{ }
e In the basis can write
or
{ }
e
'
U
y U y U Ax U AU x
+
' '
= = =
y Ax
'
' '
=
A U AU
+
'
=
A
'
U
1
A U AU
÷
'
=
Change to new orthonormal basis using
or
with the matrix given by
Because is unitary
Copyright – Michael D. Fayer, 2007
1
A U AU
÷
'
=
Extremely Important
Can change the matrix representing an operator in one orthonormal basis
into the equivalent matrix in a different orthonormal basis.
Called
Similarity Transformation
Copyright – Michael D. Fayer, 2007
y Ax AB C A B C = = + =
y A x A B C A B C
' ' ' ' ' ' '
' '
= = + =
In basis
{ }
e
Go into basis
{ }
e
'
Relations unchanged by change of basis.
AB C =
U ABU UCU
+ +
=
U U
+
AB
U AU U BU UCU
+ + +
=
A B C
' ' '
=
Example
Can insert between because
Therefore
1
1 U U U U
+ ÷
= =
Copyright – Michael D. Fayer, 2007
Isomorphism between operators in abstract vector space
and matrix representatives.
Because of isomorphism not necessary to distinguish
abstract vectors and operators
from their matrix representatives.
The matrices (for operators) and the representatives (for vectors)
can be used in place of the real things.
Copyright – Michael D. Fayer, 2007
Hermitian Operators and Matrices
Hermitian operator
x A y y A x =
Hermitian operator Hermitian Matrix
A A
+
=
+  complex conjugate transpose  Hermitian conjugate
Copyright – Michael D. Fayer, 2007
1 2
,
N
U U U
1
2
0 0
0 0
0
0
N
A
o
o
o
 


'
=


\ .
A U U o =
Theorem (Proof: Powell and Craseman, P. 303 – 307, or linear algebra book)
For a Hermitian operator A in a linear vector space of N dimensions,
there exists an orthonormal basis,
relative to which A is represented by a diagonal matrix
.
The vectors, , and the corresponding real numbers, o
i
, are the
solutions of the Eigenvalue Equation
and there are no others.
i
U
Copyright – Michael D. Fayer, 2007
Application of Theorem
Operator A represented by matrix
in some basis . The basis is any convenient basis.
In general, the matrix will not be diagonal.
A
{ }
i
e
There exists some new basis eigenvectors
{ }
i
U
in which represents operator and is diagonal eigenvalues.
A
'
To get from to
unitary transformation.
{ } { }
.
i i
U U e =
{ }
i
e
{ }
i
U
1
A U AU
÷
'
=
Similarity transformation takes matrix in arbitrary basis
into diagonal matrix with eigenvalues on the diagonal.
Copyright – Michael D. Fayer, 2007
Matrices and Q.M.
Previously represented state of system by vector in abstract vector space.
Dynamical variables represented by linear operators.
Operators produce linear transformations.
Real dynamical variables (observables) are represented by Hermitian operators.
Observables are eigenvalues of Hermitian operators.
Solution of eigenvalue problem gives eigenvalues and eigenvectors.
y A x =
A S S o =
Copyright – Michael D. Fayer, 2007
Matrix Representation
Hermitian operators replaced by Hermitian matrix representations.
In proper basis, is the diagonalized Hermitian matrix and
the diagonal matrix elements are the eigenvalues (observables).
A suitable transformation takes (arbitrary basis) into
(diagonal – eigenvector basis)
Diagonalization of matrix gives eigenvalues and eigenvectors.
Matrix formulation is another way of dealing with operators
and solving eigenvalue problems.
A A ÷
A
'
1
U AU
÷
A
A
'
1
. A U AU
÷
'
=
Copyright – Michael D. Fayer, 2007
All rules about kets, operators, etc. still apply.
Example
Two Hermitian matrices
can be simultaneously diagonalized by the same unitary
transformation if and only if they commute.
and A B
All ideas about matrices also true for infinite dimensional matrices.
Copyright – Michael D. Fayer, 2007
Example – Harmonic Oscillator
Have already solved – use occupation number representation kets and bras
(already diagonal).
( )
2 2
1
2
H P x = +
( )
1
2
aa a a
+ +
= +
1 a n n n = ÷
1 1 a n n n
+
= + +
0 1 2 3 · · · ·
0
1
2
3
a
=
·
·
·
·
0 1 0 0 0
0 0 2 0 0
0 0 0 3 0
0 0 0 0 4
 
· · ·






 · ·

· ·


· ·


·
\ .
0 0 0
0 1 1
0 2 0
1 0 0
1 1 0
1 2 2
1 3 0
a
a
a
a
a
a
a
=
=
=
=
=
=
=
matrix elements of a
Copyright – Michael D. Fayer, 2007
0 0 0 0
1 0 0 0
0 2 0 0
0 0 3 0
0 0 0
4
a
+
 
· ·

· ·


· ·

=
· ·



· ·

· · · ·
· ·
\ .
( )
1
2
H aa a a
+ +
= +
0 1 0 0
0 0 2 0
0 0 0 3
4
0 0 0 0
aa
+
 
·
·

·
·


·
·

=
·


 · · · · · ·

· · · · · ·
\ .
0 0 0 0
1 0 0 0
0 2 0 0
0 0 3 0
0 0 0
4
 
· ·

· ·


· ·

· ·



· ·

· · · ·
· ·
\ .
1 0 0 0
0 2 0 0
0 0 3 0
0 0 0 4
·
 

·

 = ·

·


· · · · ·
\ .
Copyright – Michael D. Fayer, 2007
0 0 0 0
1 0 0 0
0 2 0 0
0 0 3 0
0 0
0 4
a a
+
 
· · ·

· · ·


· · ·

=
· · ·



· · ·

· ·
· · · · · \ .
0 1 0 0
0 0 2 0
0 0 0 3
4
0 0 0 0
 
·
·

·
·


·
·

·


 · · · · · ·

· · · · · ·
\ .
0 0 0 0
0 1 0 0
0 0 2 0
0 0 0 3
·
 

·

 = ·

·


· · · · ·
\ .
( )
1
2
H aa a a
+ +
= +
Copyright – Michael D. Fayer, 2007
aa
+
a a
+
1 0 0 0 1 2 0 0 0
0 3 0 0 0 3 2 0 0
1
0 0 5 0 0 0 5 2 0
2
0 0 0 7 0 0 0 7 2
H
· ·
   
 
· ·
 
  = = · ·
 
· ·
 
 
· · · · · · · · · ·
\ . \ .
H Adding the matrices and and multiplying by ½ gives
The matrix is diagonal with eigenvalues on diagonal. In normal units
the matrix would be multiplied by . e
This example shows idea, but not how to diagonalize matrix when you
don’t already know the eigenvectors.
Copyright – Michael D. Fayer, 2007
Diagonalization
Eigenvalue equation
u Au o =
matrix representing
operator
representative of
eigenvector
eigenvalue
0 u Au o ÷ =
( ) ( )
1
0 1, 2
N
ij ij j
j
a u i N o o
=
÷ = =
¿
In terms of the components
( )
( )
( )
11 1 12 2 13 3
21 1 22 2 23 3
31 1 32 2 33 3
0
0
0
a u a u a u
a u a u a u
a u a u a u
o
o
o
÷ + + + =
+ ÷ + + =
+ + ÷ + =
This represents a system of equations
We know the a
ij
.
We don't know
o  the eigenvalues
u
i
 the vector representatives,
one for each eigenvalue.
Copyright – Michael D. Fayer, 2007
Besides the trivial solution
1 2
0
N
u u u = = =
( )
( )
( )
11 12 13
21 22 23
31 32 33
0
a a a
a a a
a a a
o
o
o
÷ · · ·
÷ · · ·
÷
=
· ·
· ·
· ·
A solution only exists if the determinant of the coefficients of the u
i
vanishes.
Expanding the determinant gives N
th
degree equation for the
the unknown o's (eigenvalues).
* * *
1 1 2 2
1
N N
u u u u u u + +· · · + =
Then substituting one eigenvalue at a time into system of equations,
the u
i
(eigenvector representatives) are found.
N equations for u's gives only N  1 conditions.
Use normalization.
know a
ij
, don't know o's
Copyright – Michael D. Fayer, 2007
Example  Degenerate Two State Problem
Basis  time independent kets orthonormal.
o 
0
H E o o ¸  = +
0
H E   ¸ o = +
o and  not eigenkets.
Coupling ¸.
These equations define H.
0
0
H E
H
H
H E
o o
 o ¸
o  ¸
 
=
=
=
=
The matrix elements are
0
0
E
H
E
o 
o ¸
 ¸
 
=

\ .
And the Hamiltonian matrix is
Copyright – Michael D. Fayer, 2007
0
0
( ) 0
( ) 0
E
E
ì o ¸ 
¸ o ì 
÷ + =
+ ÷ =
The corresponding system of equations is
0
0
0
E
E
ì ¸
¸ ì
÷
=
÷
These only have a solution if the determinant of the coefficients vanish.
Ground State
E = 0
E
0
2¸
Excited State
Dimer Splitting
Copyright – Michael D. Fayer, 2007
( )
2
2
0
0 E ì ¸ ÷ ÷ =
Expanding
0
E ì ¸
+
= +
0
E ì ¸
÷
= ÷
Energy Eigenvalues
2 2 2
0 0
2 0 E E ì ì ¸ ÷ + ÷ =
a b
a b
o 
o 
+ +
÷ ÷
+ = +
÷ = +
+ ÷
 
, a b
+ +  
, a b
÷ ÷
To obtain Eigenvectors
Use system of equations for each eigenvalue.
Eigenvectors associated with ì
+
and ì

.
and are the vector representatives of and
in the
, basis set. o 
We want to find these.
Copyright – Michael D. Fayer, 2007
0
E ì ¸
+
= +
11 12
( ) 0 H a H b ì
+ + +
÷ + =
21 22
( ) 0 H a H b ì
+ + +
+ ÷ =
First, for the
eigenvalue
write system of equations.
0 a b ¸ ¸
+ +
÷ + =
0 a b ¸ ¸
+ +
÷ =
The result is
H
11 12 21 22
; ; ; H H H H H H H H
oo o o 
= = = = Matrix elements of
0
0
H E
H
H
H E
o o
 o ¸
o  ¸
 
=
=
=
=
The matrix elements are
Copyright – Michael D. Fayer, 2007
0 0
0 ( ) E E a b ¸ ¸
+ +
÷ ÷ + =
0 0
0 ( ) a E E b ¸ ¸
+ +
+ ÷ ÷ =
An equivalent way to get the equations is to use a matrix form.
0
0
0
0
E a
E b
ì ¸
¸ ì
+ +
+ +
÷     
=
  
÷
\ . \ . \ .
Substitute
0
E ì ¸
+
= +
0 0
0 0
0
0
E E a
E E b
¸ ¸
¸ ¸
+
+
÷ ÷     
=
  
÷ ÷
\ . \ . \ .
0
0
a
b
¸ ¸
¸ ¸
+
+
÷      
=
  
÷
\ . \ . \ .
Multiplying the matrix by the column vector representative gives equations.
0 a b ¸ ¸
+ +
÷ + =
0 a b ¸ ¸
+ +
÷ =
Copyright – Michael D. Fayer, 2007
0 a b ¸ ¸
+ +
÷ + =
0 a b ¸ ¸
+ +
÷ =
The two equations are identical.
a b
+ +
=
Always get N – 1 conditions for the N unknown components.
Normalization condition gives necessary additional equation.
2 2
1 a b
+ +
+ =
1
2
a b
+ +
= =
1 1
2 2
o  + = +
Then
and
Eigenvector in terms of the
basis set.
o 
Copyright – Michael D. Fayer, 2007
For the eigenvalue
0
E ì ¸
÷
= ÷
using the matrix form to write out the equations
0
0
0
0
E a
E b
ì ¸
¸ ì
÷ ÷
÷ ÷
÷
    
=
  
÷
\ . \ . \ .
Substituting
0
E ì ¸
÷
= ÷
0
0
a
b
¸ ¸
¸ ¸
÷
÷
     
=
  
\ . \ . \ .
0 a b ¸ ¸
÷ ÷
+ =
0 a b ¸ ¸
÷ ÷
+ =
a b
÷ ÷
= ÷
1 1
2 2
a b
÷ ÷
= = ÷
1 1
2 2
o  ÷ = ÷
These equations give
Using normalization
Therefore
Copyright – Michael D. Fayer, 2007
1
H U HU
÷
'
=
Can diagonalize by transformation
diagonal not diagonal
Transformation matrix consists of representatives of eigenvectors
in original basis.
1/ 2 1/ 2
1/ 2 1/ 2
a a
U
b b
+ ÷
+ ÷
 
 
= =



÷ \ .
\ .
1
1/ 2 1/ 2
1/ 2 1/ 2
U
÷
 
=


÷
\ .
complex conjugate transpose
Copyright – Michael D. Fayer, 2007
0
0
1/ 2 1/ 2 1/ 2 1/ 2
1/ 2 1/ 2 1/ 2 1/ 2
E
H
E
¸
¸
   
 
'
=
 

 
÷ ÷ \ .
\ . \ .
Then
0
0
1 1 1 1
1
2 1 1 1 1
E
H
E
¸
¸
     
'
=
  
÷ ÷
\ . \ . \ .
1/ 2
Factoring out
0 0
0 0
1 1
1
2 1 1
E E
H
E E
¸ ¸
¸ ¸
+ ÷    
'
=
 
+ ÷ + ÷
\ .\ .
after matrix multiplication
0
0
0
0
E
H
E
¸
¸
+  
'
=

÷
\ .
more matrix multiplication
diagonal with eigenvalues on diagonal
Copyright – Michael D. Fayer, 2007